Re: [AD] Using memmove in blit()?

[aj <aj@xxxxxxxxxx>]

> Anyone else care to do some benchmarks,

I'll do a few tests with memmove()
I vaguely think i have done this before, but find no mention of it in my 
current code base; therefore must have come to the same conclusion that 
it was of little difference to allegros current code...  nevertheless i 
shall try it out again.

I've also had some success with SSE2 for mem->mem blit, and the results are:
allegro's blit = 51,569 ticks
SSE2, (no unrolling, no-pollute-cache instruction) = 29,377 ticks
SSE2, (unrolling*4, no-pollute-cache instruction) = 28,492 ticks

I also see opportunities for mem->mem  masked_blit()

But SSE2 can't happen until there is a mechanism for using 
aligned_malloc() and aligned_free()